I have developed a chatbot using Amazon Lex integrated with the bedrock knowledge base, employing the anthropic Claude v2 model. The bot features three intents: a default FallbackIntent, a custom intent for greetings, and the built-in AMAZON.QnAIntent for handling Q&A queries.
While the bot functions correctly in providing accurate responses based on the data stored in the knowledge base, I have encountered an issue with the analytics feature. Upon analyzing the conversations, I consistently observe that the bot invokes the FallbackIntent instead of the expected AMAZON.QnAIntent, despite providing correct answers derived from the Q&A intent.
This behavior is perplexing as the bot's functionality aligns with the Q&A intent's purpose, yet the analytics reflect otherwise. I seek assistance in resolving this discrepancy to ensure accurate tracking of intent invocations in the chatbot's analytics.
AWS OFFICIALAktualisiert vor 2 Jahren